AI安全论坛将于8月8日在拉斯维加斯举办，旨在汇集研究人员、工程师和政策制定者，共同加速AI安全发展。论坛将重点关注AI安全现状、人才培养和合作共赢，旨在防止AI带来的潜在风险，确保AI安全发展。

🎯 **AI安全现状与挑战:** 论坛将探讨AI安全领域面临的挑战，例如模型权重安全、硬件安全、对抗攻击等问题。来自Anthropic、RAND、Pattern Labs等机构的专家将分享他们在AI安全方面的经验和见解。

🎯 **人才培养与合作:** 论坛旨在吸引来自AI实验室、学术机构、政府机构和科技公司的专家参与，共同推动AI安全人才培养和合作。论坛将为参与者提供交流学习的机会，促进跨领域合作，共同解决AI安全问题。

🎯 **未来展望与安全 implications:** 论坛将探讨未来AI发展趋势及其对安全的影响，例如通用人工智能 (AGI) 的发展和安全问题。来自Google DeepMind、IFP等机构的专家将分享他们的观点和研究成果，为未来AI安全发展指明方向。

🎯 **多元参与者群体:** 论坛将吸引来自不同领域的专家参与，包括AI实验室、学术机构、政府机构、科技公司、独立研究人员、咨询机构和非营利组织等。多元的参与者群体将为论坛带来丰富的视角和经验，促进更全面的讨论和合作。

🎯 **共同目标：防止全球灾难:** 论坛的核心目标是确保AI安全发展，防止AI带来的潜在风险，确保AI技术造福人类。论坛将通过建立共同知识、培养人才、促进合作等方式，加速AI安全发展，为构建一个安全可靠的AI未来而努力。

Published on July 12, 2024 11:01 PM GMT

Jeffrey Ladish, Caleb Parikh, and I are running the AI security forum, a 1-day event on Thursday, August 8th '24 in Las Vegas, the day before DEFCON. This is a continuation of last year's X-infosec Forum.

We thought that many of you reading this might be interested in attending. If you're interested, please apply here by July 20th. If you can’t attend but have recommendations for other people to invite, please let us know here.

We think that there’s a significant chance transformative AI will be developed within the next 10 years, and that securing those systems is on the critical path to preventing a global catastrophe. It’s not clear that we’re on track to do that in time, so we are convening researchers, engineers, and policymakers with the goal of significantly accelerating AI security. We're aiming to do that by establishing common knowledge about the state of AI security and AI progress, mobilising talent to fill the most pressing gaps, and fostering collaborations.

Keynote speakers:

Jason Clinton, CISO @ Anthrophic, AI lab security progress and challengesSella Nevo, Director Meselson Center @ RAND, Securing model weights report: Gaps and work to be doneDan Lahav, CEO @ Pattern Labs, Securing model weights report: Gaps and work to be doneLeopold Aschenbrenner, Forecasting AGI developments and security implicationsTim Fist, Senior Technology Fellow @ IFP, Hardware security challenges for compute governanceMikel Rodriguez, AI Red Teaming Lead @ Google Deepmind, Evaluating cybersecurity capabilities of frontier models

We're inviting a similar group of people to last year, which had (roughly):

~30-40% from AI labs and research orgs (e.g., Anthropic, OpenAI, Google DeepMind) ~20-25% from academic institutions (e.g., Harvard, MIT, Stanford) ~15-20% from government agencies and think tanks (e.g., CISA, RAND) ~15% from other tech companies (e.g., Google, Microsoft, Intel) ~10-15% independent researchers, consultants, and representatives from nonprofits and funding organizations.

Please see our website for more information. Thanks!

Discuss